REAL_OPTLAY:internal error

[SKM]

hi
thanks for quick reply.
i did Gamma point only for the same input files. This time even ENCUT 616eV did not run successfully. but the error is different. ERROR: SBESSELITER : nicht konvergent and even the elapsed time is ticking, but the log file did not show any output after few basic lines of output.

ENCUT 617eV, same error as before.
attached both tar files for your check.
regards

[martin.schlipf]

I tried the Gamma-only calculation with Vasp 5.4.4 and Vasp 6.1 and could not reproduce the "ERROR: SBESSELITER" issue. However, while Vasp 5.4.4 does indeed crash with a REAL_OPTLAY error, this error was apparently fixed as Vasp 6.1. Perhaps someone else knows of a workaround this issue for Vasp 5.4.4, otherwise I can only suggest to upgrade to the current version of Vasp.

[SKM]

hi
Thanks for quick reply. I Am not clear what did you get. As there are now Version 5.4.4 and two inputs ENCUT 616 and 617 eV. Now what did you get with what combination of runs?

I try to ask our System Admin, if they know the fix or ask for Version 6.

Regards

[martin.schlipf]

Upon startup, I experience a crash when using Vasp.5.4.4 and the e617 example (REAL_OPTLAY). There is no crash at startup for the e616 example. Using Vasp.6 neither test crashes at the startup.

[SKM]

Thanks for the clarification.
i run the model again in Version 6.1.0. it runs now without such errors for ENCUTs 616 and 617eV test cases.
but all the runs shown a warning message as

| The determination of the symmetry of your systems shows a strong |
| dependence on the tolerance parameter SYMPREC. This is often the |
| result of "noise" in the structural parameters (lattice vectors |
| and/or ionic positions). Set NWRITE=3 for more details.

but I Am not sure how much effect this error will show on the results?

However, the ENCUT convergence test is still in the dilemma that why the total energy is not stabilised. I will post it separately under relevant topic or a new one.

Regards

[martin.schlipf]

This warning appears if your system is close to a structure with higher symmetry. So depending on the tolerance you set for the determination of the symmetry you'll get either the low or the high symmetry structure. Depending on what you want to do this may cause errors: On the one hand, calculating the high symmetry structure may cause degeneracies between states, which you do not want if you really wanted the low symmetry structure. On the other hand, higher symmetry typically means a more numerically stable and faster calculation. To resolve this you can either symmetrize your structure (to guarantee the high symmetry structure) or break the symmetry further (to guarantee the low symmetry structure).